ajmoss
29th June 2002, 20:18
I have tried numerous times to elicit a recommendation as to the bit rate I should use in DivX 5. Each time I've been told that there is no single answer to this, and I should experiment. The thing is, I really don't have time to encode each of my files several times over, with parameters varying by 10% or less, and my powers of observation wouldn't be good enough to tell the difference anyway.
In fact it goes further than that. My computer is fast enough to capture and encode into DivX in real time, and it would be a huge waste of disk space and CPU time to perform an initial capture into the Huffy codec. This means I only get one chance to set the bit rate correctly, and furthermore I can't use two pass encoding.
In the files http://www.doom9.org/gknot-basics-divx3.htm and http://www.doom9.org/gknot-basics.htm (the latter is for DivX 4), the advice given is:
It's a bit guesswork as you cannot predict compressibility of your movie but here's some hints:
It's not a good idea to go below 0.20 bits/pixel*frame
If you go above ~0.35 yo might not be able to reach your desired filesize
If you go for 1 CD stay close to 0.20
If you go for 2 CDs stay around 0.27
If TV bits/pixel*frame is below ~0.13 you should raise your bitrate (and go for more CDs).
The recommendations in http://www.doom9.org/gknot-basics-divx5.htm are a bit lower, because of the greater compressability of B frames:
It's a bit guesswork as you cannot predict compressibility of your movie but here's some hints:
It's not a good idea to go much below 0.17 bits/pixel*frame
If you go for 1 CD stay close to 0.17
If you go for 2 CDs stay around 0.25
If TV bits/pixel*frame is below ~0.13 you should raise your bitrate (and go for more CDs).
This appears to be the Philosopher's Stone, the big secret that noone wants to divulge.
To put it simply, I'm already familiar with the notion of a range of recommended bit rates - with MPEG-1, the bit rate of VCDs is reckoned barely adequate for the job, and so a bit rate of 1.5Mbit/s is considered advisable. ATI recommends using 2.4Mbit/s, to ensure that the limiting factor is always the limitations of the codec, not that of the bit rate.
One implication of these quotes is that the recommended bit rate scales linearly with the number of pixels. When capturing a PAL TV programme, it makes some sense to do so at half PAL resolution (352x288 pixels, the same as a VCD): the colour information is stored at this lower resolution, and it decreases the workload on the CPU running the codec. It also reduces the necessary video bit rate by three quarters.
352x288 pixels is just over 100,000; this makes 2.5 million pixels per second. By this calculation, you will get a good quality picture with B frames enabled (around 0.25 bits/pixel) at 640 kbit/s. If you add to this a good quality MP3 audio at 160 kbit/s, you get a round 800kbit/s in total.
This is probably less than you would normally use for a DVD conversion, because a DVD conversion would typically be done at a higher resolution than 352x288 pixels.
The capacity of an 80 minute CD is
8 bits * 2048 bytes * 75 frames * 60 seconds * 80 minutes
= 16,384 * 360,000
One hour's recording on a CD therefore works out at 1,638,400 bit/s.
At a bit rate of 800kbit/s, you will get two hours onto a CD.
You could safely reduce the video and audio bit rates by 20%, to 500kbit/s video (0.20 bits/pixel) and 128kbit/s audio. This will give you 2.5 hours recording time on a CD. If you need to squeeze on any more than that, you really need to reduce the video resolution.
An hour-long British programme on Channel 4 (in the UK) is typically 50 minutes without the adverts. You could store this at a resolution of 800kbit/s video plus 160kbit/s audio, or if you like, 768kbit/s video plus 192kbit/s audio. At this bit rate, you get 100 minutes on a CD.
An hour-long American programme, such as Buffy or the X Files, is typically only 42-43 minutes long without the adverts. At 600kbit/s video plus 160kbit/s audio, you can get three episodes onto one CD.
In fact it goes further than that. My computer is fast enough to capture and encode into DivX in real time, and it would be a huge waste of disk space and CPU time to perform an initial capture into the Huffy codec. This means I only get one chance to set the bit rate correctly, and furthermore I can't use two pass encoding.
In the files http://www.doom9.org/gknot-basics-divx3.htm and http://www.doom9.org/gknot-basics.htm (the latter is for DivX 4), the advice given is:
It's a bit guesswork as you cannot predict compressibility of your movie but here's some hints:
It's not a good idea to go below 0.20 bits/pixel*frame
If you go above ~0.35 yo might not be able to reach your desired filesize
If you go for 1 CD stay close to 0.20
If you go for 2 CDs stay around 0.27
If TV bits/pixel*frame is below ~0.13 you should raise your bitrate (and go for more CDs).
The recommendations in http://www.doom9.org/gknot-basics-divx5.htm are a bit lower, because of the greater compressability of B frames:
It's a bit guesswork as you cannot predict compressibility of your movie but here's some hints:
It's not a good idea to go much below 0.17 bits/pixel*frame
If you go for 1 CD stay close to 0.17
If you go for 2 CDs stay around 0.25
If TV bits/pixel*frame is below ~0.13 you should raise your bitrate (and go for more CDs).
This appears to be the Philosopher's Stone, the big secret that noone wants to divulge.
To put it simply, I'm already familiar with the notion of a range of recommended bit rates - with MPEG-1, the bit rate of VCDs is reckoned barely adequate for the job, and so a bit rate of 1.5Mbit/s is considered advisable. ATI recommends using 2.4Mbit/s, to ensure that the limiting factor is always the limitations of the codec, not that of the bit rate.
One implication of these quotes is that the recommended bit rate scales linearly with the number of pixels. When capturing a PAL TV programme, it makes some sense to do so at half PAL resolution (352x288 pixels, the same as a VCD): the colour information is stored at this lower resolution, and it decreases the workload on the CPU running the codec. It also reduces the necessary video bit rate by three quarters.
352x288 pixels is just over 100,000; this makes 2.5 million pixels per second. By this calculation, you will get a good quality picture with B frames enabled (around 0.25 bits/pixel) at 640 kbit/s. If you add to this a good quality MP3 audio at 160 kbit/s, you get a round 800kbit/s in total.
This is probably less than you would normally use for a DVD conversion, because a DVD conversion would typically be done at a higher resolution than 352x288 pixels.
The capacity of an 80 minute CD is
8 bits * 2048 bytes * 75 frames * 60 seconds * 80 minutes
= 16,384 * 360,000
One hour's recording on a CD therefore works out at 1,638,400 bit/s.
At a bit rate of 800kbit/s, you will get two hours onto a CD.
You could safely reduce the video and audio bit rates by 20%, to 500kbit/s video (0.20 bits/pixel) and 128kbit/s audio. This will give you 2.5 hours recording time on a CD. If you need to squeeze on any more than that, you really need to reduce the video resolution.
An hour-long British programme on Channel 4 (in the UK) is typically 50 minutes without the adverts. You could store this at a resolution of 800kbit/s video plus 160kbit/s audio, or if you like, 768kbit/s video plus 192kbit/s audio. At this bit rate, you get 100 minutes on a CD.
An hour-long American programme, such as Buffy or the X Files, is typically only 42-43 minutes long without the adverts. At 600kbit/s video plus 160kbit/s audio, you can get three episodes onto one CD.